[Perlfect-search] update the index?

[Daniel Naber]

On Sam, 12 Aug 2000, you wrote:

> 3. We could cluster the whole collection of documents(not only the
> search results) and the let the user find "similar document". So then

Actually I think a link called "Search for similar documents" is not very 
usful, it just doesn't offer enough information in what way these 
documents will be similar. Also you normally have to see the page before 
you can decide you want similar documents. This means that the link should 
be on the page that was found, which is indeed possible: Some search 
engines load the page for you and a link the the first occurence of the 
term (I think Google does that, too).

> Imagine how useful these features could be even for a mail client.

Accessing DB_File hashes written by Perl didn't work, I just tried. Using 
db_dump I get "database needs an update" or so, and after the update I can 
dump the file, but I cannot see the values. Also db_verify gives errors. 
As there is no Perl mail client (well, at least I don't use one :-) 
everthing has to be rewritten in C++ ... which is a bit more complicated 
that just "tie %hash", it seems...

Regards
 Daniel